Information security is just as old as Information Technology itself. As soon as organizations began to depend on IT systems to run their business processes and to store and process business information, it has become necessary to protect these systems from malicious attacks. First concepts of tools for detecting and fighting off intrusions into computer networks were developed in early 1980s, and in the following three decades security analytics has evolved through several different approaches, reflecting the evolution of IT landscape as well as changing business requirements.
First-generation security tools – firewalls and intrusion detection and prevention systems (IDS/IPS) – have essentially been solutions for perimeter protection. Firewalls were traditionally deployed on the edge of a trusted internal network and were meant to prevent attacks from the outside world. First firewalls were simply packet filters that were effective for blocking known types of malicious traffic or protecting from known weaknesses in network services. Later generation of application firewalls can understand certain application layer protocols and thus provide additional protection for specific applications: mitigate cross-site scripting attacks on websites, protect databases from SQL injections, perform DLP functions, etc. Intrusion detection systems can be deployed within networks, but old signature-based systems were only capable of reliably detecting known threats and later statistical anomaly-based solutions were known to generate an overwhelming number of false alerts. In general, tuning an IDS for a specific network was always a difficult and time-consuming process.
These traditional tools are still widely deployed by many organizations and in certain scenarios serve as a useful part of enterprise security infrastructures, but recent trends in the IT industry have largely made them obsolete. Continued deperimeterization of corporate networks because of adoption of cloud and mobile services, as well as emergence of many new legitimate communication channels with external partners has made the task of protecting sensitive corporate information more and more difficult. The focus of information security has gradually shifted from perimeter protection towards detection and defense against threats within corporate networks.
The so-called Advanced Persistent Threats usually involve multiple attack vectors and consist of several covert stages. These attacks may go on undetected for months and cause significant damage for unsuspecting organizations. Often they are first uncovered by external parties, adding reputation damage to financial losses. A well-planned APT may exploit several different vulnerabilities within the organization: an unprotected gateway, a bug in an outdated application, a Zero-Day attack exploiting a previously unknown vulnerability and even social engineering, targeting the human factor often neglected by IT security.
By the mid-2000s, it was obvious that efficient detection and defense against these attacks requires a completely new approach towards network security. The need to analyze and correlate security incidents from multiple sources, to manage a large number of alerts and to be able to perform forensic analysis has led to development of a new organizational concept of Security Operations Center (SOC). An SOC is a single location where a team of experts is monitoring security-related events of entire enterprise information systems and taking actions against detected threats. Many large enterprises have established their own SOCs and for smaller organizations that cannot afford considerable investments and maintaining a skilled security staff on their own, such services are usually offered as a Managed Security Service.
The underlying technological platform of a security operations center is SIEM: Security Information and Event Management – a set of software and services for gathering, analyzing and presenting information from various sources, such as network devices, applications, logging systems, or external intelligence sources. The term has been coined in 2005 and the concept has been quickly adopted by the market: currently there are over 60 vendors offering SIEM solutions in various forms. There was a lot of initial hype around the SIEM concept, as it was offered as a turnkey solution for all security-related problems mentioned above. The reality, however, has shown that, although SIEM solutions are very capable sets of tools for data aggregation, retention and correlation, as well as for monitoring, alerting and reporting of security incidents, they are still just tools, requiring a team of experts to deploy and customize and another team to run it on daily basis.
Although SIEM solutions are currently widely adopted by most large enterprises, there are several major challenges that, according to many information security officers, are preventing them from efficiently using them:
Another common shortcoming of current SIEM solutions is lack of flexibility when dealing with unstructured data. Since many of the products are based on relational databases, they enforce applying rigid schemas to collected information and do not scale well when dealing with large amounts of data. This obviously prevents them from efficiently detecting threats in real time.
- Current SIEM solutions require specially trained security operations experts to operate; many organizations simply do not have enough resources to maintain such teams.
- Current SIEM solutions generate too many false positive alerts, forcing security teams to deal with overwhelming amounts of unnecessary information. Obviously, current correlation and anomaly detection algorithms are not efficient enough.
- The degree of integration offered by current SIEM solutions is still insufficient to provide a truly single management console for all kinds of operations. Responding to a security incident may still require performing too many separate actions using different tools.
Over the last couple of years, these challenges have led to the emergence of the “next-generation SIEM” or rather a completely new technology called Real-time Security Intelligence (RTSI). Although the market is still in its early stage, it is already possible to summarize the key differentiators of RTSI offerings from previous-generation SIEM tools:
- Real-time or near real-time detection of threats that enables quick remediation before damage is done;
- Possibility to correlate real-time and historical data from various sources, as well as apply intelligence from external security information services, thus detecting malicious operations as whole events, not separate alerts;
- Small number of clearly actionable alarms by reducing the false positive rate, as well as introducing different risk levels for incidents;
- Automated workflows for responding to detected threats, such as, for example, disrupting clearly identified malware attacks or submitting a suspicious event to a managed security service for further analysis.
The biggest technological breakthrough that made these solutions possible is Big Data analytics. The industry has finally reached the point, when business intelligence algorithms for large-scale data processing, previously affordable only to large corporations, have become commoditized. Utilizing readily available frameworks such as Apache Hadoop and inexpensive hardware, vendors are now able to build solutions for collecting, storing and analyzing huge amounts of unstructured data in real-time.
This makes it possible to combine real-time and historical analysis and identify new incidents as being related to others that occurred in the past. Combined with external security intelligence sources that provide current information about the newest vulnerabilities, this can greatly facilitate identification of ongoing APT attacks on the network. Having a large amount of historical data at hand also significantly simplifies initial calibration to the normal patterns of activity of a given network, which are then used to identify anomalies. Existing RTSI solutions are already capable of automated calibration with very little input required from administrators.
Alerting and reporting capabilities of RTSI solutions are also significantly improved. Big Data analytics technology can generate a small number of concise and clearly categorized alerts to allow even an inexperienced person to make a relevant decision, yet provides a forensic expert with much more details about the incident and its relations with other historical anomalies.
As mentioned above, the RTSI market is still in its early stage. There are many new offerings with various scopes of functionality from both established IT security vendors as well as startups available today or planned for release in near future. It is still difficult to predict in which direction the market will evolve and which features should be expected from an innovation leader. However, it is already clear that only the vendors that will offer complete solutions and not just set of tools will win the market. It is important to understand that Real-time Security Intelligence is more than just SIEM 2.0.